1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Spatie\Sitemap; |
4
|
|
|
|
5
|
|
|
use Closure; |
6
|
|
|
use GuzzleHttp\Psr7\Uri; |
7
|
|
|
use Illuminate\Support\Collection; |
8
|
|
|
use Psr\Http\Message\ResponseInterface; |
9
|
|
|
use Psr\Http\Message\UriInterface; |
10
|
|
|
use Spatie\Crawler\Crawler; |
11
|
|
|
use Spatie\Crawler\CrawlProfile; |
12
|
|
|
use Spatie\Sitemap\Crawler\Observer; |
13
|
|
|
use Spatie\Sitemap\Crawler\Profile; |
14
|
|
|
use Spatie\Sitemap\Tags\Url; |
15
|
|
|
|
16
|
|
|
class SitemapGenerator |
17
|
|
|
{ |
18
|
|
|
/** @var \Illuminate\Support\Collection */ |
19
|
|
|
protected $sitemaps; |
20
|
|
|
|
21
|
|
|
/** @var \GuzzleHttp\Psr7\Uri */ |
22
|
|
|
protected $urlToBeCrawled = ''; |
23
|
|
|
|
24
|
|
|
/** @var \Spatie\Crawler\Crawler */ |
25
|
|
|
protected $crawler; |
26
|
|
|
|
27
|
|
|
/** @var callable */ |
28
|
|
|
protected $shouldCrawl; |
29
|
|
|
|
30
|
|
|
/** @var callable */ |
31
|
|
|
protected $hasCrawled; |
32
|
|
|
|
33
|
|
|
/** @var int */ |
34
|
|
|
protected $concurrency = 10; |
35
|
|
|
|
36
|
|
|
/** @var bool|int */ |
37
|
|
|
protected $maximumTagsPerSitemap = false; |
38
|
|
|
|
39
|
|
|
/** @var int|null */ |
40
|
|
|
protected $maximumCrawlCount = null; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* @param string $urlToBeCrawled |
44
|
|
|
* |
45
|
|
|
* @return static |
46
|
|
|
*/ |
47
|
|
|
public static function create(string $urlToBeCrawled) |
48
|
|
|
{ |
49
|
|
|
return app(static::class)->setUrl($urlToBeCrawled); |
50
|
|
|
} |
51
|
|
|
|
52
|
|
|
public function __construct(Crawler $crawler) |
53
|
|
|
{ |
54
|
|
|
$this->crawler = $crawler; |
55
|
|
|
|
56
|
|
|
$this->sitemaps = new Collection([new Sitemap]); |
57
|
|
|
|
58
|
|
|
$this->hasCrawled = function (Url $url, ResponseInterface $response = null) { |
|
|
|
|
59
|
|
|
return $url; |
60
|
|
|
}; |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
public function configureCrawler(Closure $closure): self |
64
|
|
|
{ |
65
|
|
|
call_user_func_array($closure, [$this->crawler]); |
66
|
|
|
|
67
|
|
|
return $this; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
public function setConcurrency(int $concurrency) |
71
|
|
|
{ |
72
|
|
|
$this->concurrency = $concurrency; |
73
|
|
|
|
74
|
|
|
return $this; |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
public function setMaximumCrawlCount(int $maximumCrawlCount) |
78
|
|
|
{ |
79
|
|
|
$this->maximumCrawlCount = $maximumCrawlCount; |
80
|
|
|
|
81
|
|
|
return $this; |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
public function maxTagsPerSitemap(int $maximumTagsPerSitemap = 50000): self |
85
|
|
|
{ |
86
|
|
|
$this->maximumTagsPerSitemap = $maximumTagsPerSitemap; |
87
|
|
|
|
88
|
|
|
return $this; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
public function setUrl(string $urlToBeCrawled) |
92
|
|
|
{ |
93
|
|
|
$this->urlToBeCrawled = new Uri($urlToBeCrawled); |
94
|
|
|
|
95
|
|
|
if ($this->urlToBeCrawled->getPath() === '') { |
96
|
|
|
$this->urlToBeCrawled = $this->urlToBeCrawled->withPath('/'); |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
return $this; |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
public function shouldCrawl(callable $shouldCrawl) |
103
|
|
|
{ |
104
|
|
|
$this->shouldCrawl = $shouldCrawl; |
105
|
|
|
|
106
|
|
|
return $this; |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
public function hasCrawled(callable $hasCrawled) |
110
|
|
|
{ |
111
|
|
|
$this->hasCrawled = $hasCrawled; |
112
|
|
|
|
113
|
|
|
return $this; |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
public function getSitemap(): Sitemap |
117
|
|
|
{ |
118
|
|
|
if (config('sitemap.execute_javascript')) { |
119
|
|
|
$this->crawler->executeJavaScript(config('sitemap.chrome_binary_path')); |
|
|
|
|
120
|
|
|
} |
121
|
|
|
|
122
|
|
|
if (! is_null($this->maximumCrawlCount)) { |
123
|
|
|
$this->crawler->setMaximumCrawlCount($this->maximumCrawlCount); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
$this->crawler |
127
|
|
|
->setCrawlProfile($this->getCrawlProfile()) |
128
|
|
|
->setCrawlObserver($this->getCrawlObserver()) |
129
|
|
|
->setConcurrency($this->concurrency) |
130
|
|
|
->startCrawling($this->urlToBeCrawled); |
131
|
|
|
|
132
|
|
|
return $this->sitemaps->first(); |
133
|
|
|
} |
134
|
|
|
|
135
|
|
|
/** |
136
|
|
|
* @param string $path |
137
|
|
|
* |
138
|
|
|
* @return $this |
139
|
|
|
*/ |
140
|
|
|
public function writeToFile(string $path) |
141
|
|
|
{ |
142
|
|
|
$sitemap = $this->getSitemap(); |
143
|
|
|
|
144
|
|
|
if ($this->maximumTagsPerSitemap) { |
145
|
|
|
$sitemap = SitemapIndex::create(); |
146
|
|
|
$format = str_replace('.xml', '_%d.xml', $path); |
147
|
|
|
|
148
|
|
|
// Parses each sub-sitemaps, writes and pushs them into the sitemap |
149
|
|
|
// index |
150
|
|
|
$this->sitemaps->each(function (Sitemap $item, int $key) use ($sitemap, $format) { |
151
|
|
|
$path = sprintf($format, $key); |
152
|
|
|
|
153
|
|
|
$item->writeToFile(sprintf($format, $key)); |
154
|
|
|
$sitemap->add(last(explode('public', $path))); |
155
|
|
|
}); |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
$sitemap->writeToFile($path); |
159
|
|
|
|
160
|
|
|
return $this; |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
protected function getCrawlProfile(): CrawlProfile |
164
|
|
|
{ |
165
|
|
|
$shouldCrawl = function (UriInterface $url) { |
166
|
|
|
if ($url->getHost() !== $this->urlToBeCrawled->getHost()) { |
167
|
|
|
return false; |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
if (! is_callable($this->shouldCrawl)) { |
171
|
|
|
return true; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
return ($this->shouldCrawl)($url); |
175
|
|
|
}; |
176
|
|
|
|
177
|
|
|
$profileClass = config('sitemap.crawl_profile', Profile::class); |
178
|
|
|
$profile = new $profileClass($this->urlToBeCrawled); |
179
|
|
|
|
180
|
|
|
if (method_exists($profile, 'shouldCrawlCallback')) { |
181
|
|
|
$profile->shouldCrawlCallback($shouldCrawl); |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
return $profile; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
protected function getCrawlObserver(): Observer |
188
|
|
|
{ |
189
|
|
|
$performAfterUrlHasBeenCrawled = function (UriInterface $crawlerUrl, ResponseInterface $response = null) { |
190
|
|
|
$sitemapUrl = ($this->hasCrawled)(Url::create((string) $crawlerUrl), $response); |
191
|
|
|
|
192
|
|
|
if ($this->shouldStartNewSitemapFile()) { |
193
|
|
|
$this->sitemaps->push(new Sitemap); |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
if ($sitemapUrl) { |
197
|
|
|
$this->sitemaps->last()->add($sitemapUrl); |
198
|
|
|
} |
199
|
|
|
}; |
200
|
|
|
|
201
|
|
|
return new Observer($performAfterUrlHasBeenCrawled); |
202
|
|
|
} |
203
|
|
|
|
204
|
|
|
protected function shouldStartNewSitemapFile(): bool |
205
|
|
|
{ |
206
|
|
|
if (! $this->maximumTagsPerSitemap) { |
207
|
|
|
return false; |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
$currentNumberOfTags = count($this->sitemaps->last()->getTags()); |
211
|
|
|
|
212
|
|
|
return $currentNumberOfTags >= $this->maximumTagsPerSitemap; |
213
|
|
|
} |
214
|
|
|
} |
215
|
|
|
|
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.